Tag
#AI security
ChatGPT Shared-Link Account Takeover via MathLive Sandbox Escape
A shared ChatGPT link ran attacker JavaScript inside the victim session through MathLive unchecked messages; OpenAI fixed it on October 26.
Oct 31, 2026
Post-Mortem: Why Autonomous AI Agents Escape Scope and How to Contain Them
A technical post-mortem on an AI agent that wandered outside its testing scope during an authoriz...
Sep 18, 2026
How AI Catches Complex Vulnerabilities: Inside Agentic Pentesting and Exploit Chaining
Discover how agentic AI catches business logic flaws rule-based scanners miss. See a real exploit...
Jul 28, 2026
The App Was Never Opened
Agentic harnesses change what an LLM can do in mobile app security testing. On its own, a model c...
Jun 25, 2026